[
https://issues.apache.org/jira/browse/HDFS-8334?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14531730#comment-14531730
]
Zhe Zhang commented on HDFS-8334:
---------------------------------
Good question.
bq. after we have the writing EC file functionality
Both striped writing and reading logics are fairly complex (as well as codec).
I think we should have true unit testing for all 3, as well as an end-to-end
testing.
One example is [described |
https://issues.apache.org/jira/browse/HDFS-7678?focusedCommentId=14531718&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-14531718]
under HDFS-7678. Also, we have never verified that parity blocks are
successfully written to DNs with correct content.
> Erasure coding: rename DFSStripedInputStream related test classes
> -----------------------------------------------------------------
>
> Key: HDFS-8334
> URL: https://issues.apache.org/jira/browse/HDFS-8334
> Project: Hadoop HDFS
> Issue Type: Sub-task
> Components: test
> Reporter: Zhe Zhang
> Assignee: Zhe Zhang
> Attachments: HDFS-8334-HDFS-7285.0.patch
>
>
> Our current TestDFSStripedInputStream is actually the end-2-end test, and
> should be named TestWriteReadStripedFile, and should eventually subclass
> TestWriteRead. Current TestReadStripedFile is actually the internal unit
> testing class for DFSStripedInputStream
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
